Tripartite Palindromic Partition of Integer (HP 50g) Challenge

[2old2randr]

By optimization, it seems that you mean converting the user RPL program to use stack operations as far as possible (instead of using local variables). Is that correct?

I compared the speed of my version of ALGO1 vs your streamlined version. The time taken to execute the algorithm on an input corresponding to the number 2345678 is 0.95-0.97 seconds vs. 0.99-1.03 seconds (on a physical 50g).

Although there is a marginal improvement in execution time (2-7%), there is a severe loss in readability especially if you are comparing the code to the algorithm in the paper so my question is - is it worth doing this?